Today, we’re going to take a few minutes and find out what is virtualization? There’s a lot of hype about this technology in the industry. Pretty much all of the major operating system vendors are virtualizing or supporting virtualization in one way or another. And, so, let’s take a look and find out what virtualization is and what it does to our IT infrastructure. And even more importantly, how it effects colocation in Tulsa, OK.
So, we have a model that we’ve used for really a couple of decades in the IT industry where we have some vendor brand hardware down here. Insert your favorite. It could be Dell, HP, IBM, whatever you run in your shop. And, then, we have various operating systems that run on top of this hardware. Now, the operating systems are bound to the hardware by drivers. So, if I have an HP ProLiant DL360 for example and I install a copy of Windows 2008 server, if I go into Device Manager, I’ll see drivers that are specific to that piece of hardware. So, I may see the SCSI board driver for that particular server or the video driver for that particular server. And then, of course, on top of this, we run various types of applications and services. Now, this is the way that things have been for a long time in the IT industry and virtualization changes all this. There are a lot of limitations with this traditional model. For example, with this traditional model, we have one operating system per server and that’s it. This operating system monopolizes and controls this piece of hardware. And, of course, the OS is bound to this particular piece of hardware by the drivers. So, of course, this makes backups, and more importantly, restores and migrations very challenging.
So, going back to our example of having a DL360, if I have an operating system on there with applications and I want to migrate that to a different server, maybe a different model HP server, I want to move it over to another vendor server so I can shut this one down and modify some hardware, well as you know, in a non-virtualized space, it’s not really that easy. We can’t just take operating systems, for the most part, and move them around to other servers. Even if we image them correctly, the operating system, when it goes to boot, it expects to see the hardware based upon the driver set that it has currently installed. Failing that, it will typically crash, or blue screen, or something like that.
So, as we look at disaster recovery and talk about disaster recovery, backups are relatively easy, but the restores are the problem. I’ve heard it said that backups are optionals, but restores are not. So, imagine that you have a server that you’ve been backing up faithfully for three or four years, and you get down there, and one day the server fries. And now you have to take that image that you have — you have your data. You have it all backed up, but you want to restore that somewhere in your environment. Well, again, you can’t just take that, for the most part, and restore it to some other server because that image is looking to have that same hardware in there supported by the embedded drivers.
Now, because of this, another side effect of this model in the industry is that we end up with lots of underutilized servers. Pretty much every application vendor wants you to have an isolated server for their application. So, if I go out and I buy an accounting package, the vendor doesn’t want me to take his accounting package and put it on a server with five or six other applications that may cause performance degradation or support problems for the vendor. It’s very difficult to support that kind of thing. And so, typically, your application vendors require that you have an isolated server dedicated to their application or task. And, so, we end up with all these underutilized servers out here in the industry. If you look at industry studies, about 5% to 15% average process utilization across the Intel market and this has become known as server sprawl. Of course, there’s a lot of wasted money here, right? If I spend $10,000 on this server and I’m only using 10% of it, well, in a way of looking at it, I’m kind of wasting $9,000 worth of computing power. Of course, another side effect, an additional cost center of this, is all of the additional support contracts that I have to have for these extra servers. The power consumption probably has extra servers. And, as we’ll see with virtualization, we’re going to be able to radically reduce a lot of these costs and simplify and improve management at the same time.
Read the rest of this entry